More Tesla investigations

[JeffKeryk]

Who says the brain isn’t checking a thousand things in an instant? When you look at something our eyes connected to the brain are checking things as fast as light hits the scene. When we touch, hear, or think it is beyond understanding. When we think of a memory how does that even work. Like houses we lived in as children. The image in the mind is just there. What is that image anyway does anyone know?

OK. Try looking 360* at once.
The point is, once a comluter and its periphrials are programmed, they will react in a known way. There are about 7B people on the earth; do any 2 react in exactly the same way?
Once a program is updated, it can be pushed out to every instance of that program.
How long does it take for, say 100M people to be updated?

Computers are better because they are predictable and accept the changes programmed to them.

 

[JeffKeryk]

“Can any mind check a thousand things in an instant?”


No but a human can make a critical decision. Like the scenario posted earlier, a plastic bag flies across the freeway. The computer detects an object and brakes. The human knows there is no harm and keeps going thus avoiding a accident happening behind them.

Over the years we will encounter scenarios like that or some that require decisive thinking. To avoid hitting another person for example forces you to hit a tree or another car.

Sure. Try and predict what that decision will be.

 

[Job]

OK. Try looking 360* at once.
The point is, once a comluter and its periphrials are programmed, they will react in a known way. There are about 7B people on the earth; do any 2 react in exactly the same way?
Once a program is updated, it can be pushed out to every instance of that program.
How long does it take for, say 100M people to be updated?

Computers are better because they are predictable and accept the changes programmed to them.

Keep your eyes on the fly with all its eyes. We are hunters we need stereoscopic front facing eyes to chase our prey. That’s what we get. Some animals have eyes on the side to watch with. Some can move their eyes more to get both needs met, side and front. Flies have fantastic eyes. Flies can learn. Not enough to stay out of the spider's web every time.
Updating is instantaneous with everything that has a brain. All 7 billion on earth react exactly the same way with vision and all the senses according to the condition of their sensors. All 7 billion sighted people with good vision update what they see in real time all the time. The brain is always updating. If the life form gets injured most of the time it can heal itself. If a oerson goes blind their other senses sharpen automatically. Updating with no help. Computers are nothing compared to any life form with a brain, or even plants without a brain. To think we can create something equal to a fly is just arrogance and lack of observation. For some reason we humans have a lot of arrogance as to what we can do.
As you read this page your brain is getting information from the eyes and creating an image. It looks real that’s how good the brain is integrated. Technically nothing is real as seen because the light waves had to travel to the eyes. That takes time. Everything is in the past we see. So it no longer exists as seen. The brain is quite something.
Now think of yourself as a child. There it is in your mind right away. You see it, I see mine, but what is this? Its not like vision where we look at pictures and real life. Good luck making a machine like this, never ever going to happen, not even close.

 

[ZeeOSix]

Autonomous driving cars do not exist.

They do ... but not in the context you're thinking (ie, used by the general public).

And the idea that drivers are better is ludacris (no offense to anyone).
Can any mind check a thousand things in an instant?

I'd bet a very good human driver could make more accurate decisions in real life situations and have more situational foresight than any "self driving/AP" cars of today can make. I'm talking about a human that is actually applying 100% of their attention to driving, like a fighter pilot in a dog fight would do.

 

[ZeeOSix]

"The brain is always updating."

Until it starts "shorting out".

 

[JeffKeryk]

They do ... but not in the context you're thinking (ie, used by the general public).


I'd bet a very good human driver could make more accurate decisions in real life situations and have more situational foresight than any "self driving/AP" cars of today can make. I'm talking about a human that is actually applying 100% of their attention to driving, like a fighter pilot in a dog fight would do.

I 100% agree that there are many situations a good driver would out perform a driver assist system.
So does Tesla; that's why they instruct you to be ready at all times.
There are also many situations where a computer is at least a valuable assist or even better than a driver.

Question: Does a fighter pilot's fighter jet use computers?

 

[Job]

OK. Try looking 360* at once.
The point is, once a comluter and its periphrials are programmed, they will react in a known way. There are about 7B people on the earth; do any 2 react in exactly the same way?
Once a program is updated, it can be pushed out to every instance of that program.
How long does it take for, say 100M people to be updated?

Computers are better because they are predictable and accept the changes programmed to them.

Also a machine doesn’t see any image or create an image at all. It takes a life form with a brain to see an image. A machine is dead, an assembly of materials. It is totally amazing how life forms have been made to interpret their surroundings as images with a brain and thought.

 

[fsdork]

Also a machine doesn’t see any image or create an image at all. It takes a life form with a brain to see an image. A machine is dead, an assembly of materials. It is totally amazing how life forms have been made to interpret their surroundings as images with a brain and thought.

You've referenced the human ability to see an image in multiple posts in this thread. Why?

There is no doubt that the best human driver is currently better than the best self-driving system in most circumstances - today. FSD doesn't need to learn, see, or understand an image as well as a human to eventually become safer and more capable. The people behind FSD will tweak the programming after each learning experience. Something went bad today? Ok, what do we need to tweak in the logic to keep that from happening next time? And so on. With each revision, there are less bad days.

People screw up a lot. Computers will do better.

 

[ecotourist]

Computers are very good at well defined repetitive activities. Humans are better at decision making in novel situations.

My Tesla created a (somewhat) dangerous situation by overriding my decision to do nothing, by vigorously braking . That braking was unnecessary and created a danger where none had previously existed. In my opinion it should not have been able to do that. And do note that I was not using an autonomous driving system. I was just driving (probably with cruise enabled).

I'm all for computers in the right situation. But humans are better at decision making in complex situations.

 

[PandaBear]

Keep your eyes on the fly with all its eyes. We are hunters we need stereoscopic front facing eyes to chase our prey. That’s what we get. Some animals have eyes on the side to watch with. Some can move their eyes more to get both needs met, side and front. Flies have fantastic eyes. Flies can learn. Not enough to stay out of the spider's web every time.
Updating is instantaneous with everything that has a brain. All 7 billion on earth react exactly the same way with vision and all the senses according to the condition of their sensors. All 7 billion sighted people with good vision update what they see in real time all the time. The brain is always updating. If the life form gets injured most of the time it can heal itself. If a oerson goes blind their other senses sharpen automatically. Updating with no help. Computers are nothing compared to any life form with a brain, or even plants without a brain. To think we can create something equal to a fly is just arrogance and lack of observation. For some reason we humans have a lot of arrogance as to what we can do.
As you read this page your brain is getting information from the eyes and creating an image. It looks real that’s how good the brain is integrated. Technically nothing is real as seen because the light waves had to travel to the eyes. That takes time. Everything is in the past we see. So it no longer exists as seen. The brain is quite something.
Now think of yourself as a child. There it is in your mind right away. You see it, I see mine, but what is this? Its not like vision where we look at pictures and real life. Good luck making a machine like this, never ever going to happen, not even close.

1) Nothing is instant in the world, not even speed of light. Not computer (i.e. 1ms response time), and certainly not human reaction time (i.e about 20ms even if you are good, 200ms if you are bad).
2) Statistics out there says human are good at many things, just not as fast as computer in repetitive things like shooting down a rocket or a fly.
3) Chip is cheap, human is not, that is the fundamental reason why we have computer and self service stuff these days. Horses were the original self driving AI, it was not as cost effective and efficient as automobile with human driver, hence we have to deal with all these accidents and drunk driving now.

 

[PandaBear]

You've referenced the human ability to see an image in multiple posts in this thread. Why?

There is no doubt that the best human driver is currently better than the best self-driving system in most circumstances - today. FSD doesn't need to learn, see, or understand an image as well as a human to eventually become safer and more capable. The people behind FSD will tweak the programming after each learning experience. Something went bad today? Ok, what do we need to tweak in the logic to keep that from happening next time? And so on. With each revision, there are less bad days.

People screw up a lot. Computers will do better.

Human has 16 years to develop the motor skill since birth before he or she can take a driving test. Self driving is "getting there", and usually good enough to be as good as a 16 year old new driver. Human has a few years to improve after passing the test before insurance is cheap, we'll likely need a decade or two to know about how self driving do based on insurance claim data.

That says, removing drunk driving and reckless driving from human would dramatically reduce accidents. We haven't put breath analyzer in human driven cars but some people here demand logs on every single driving test on self driving cars, despite them not a developer of the company and know what those logs means.

Let's be fair, I demand dash cam upload to youtube for teen drivers between 16 and 18, and breath analyzer data on each engine start for all drivers. I'm sure people here will not agree to that standard despite dramatically reducing accidents for all.

 

[ctrcbob]

I guess it’s a 1980 Citation for you then. This stuff has been in cars for a while now.


Computers are getting smarter. As an example, my iPhone alerted me about a month ago that I was becoming increasingly unsteady and my fall risk had increased to a level where I needed to be aware. Now I could just shrug it off as a nuisance alert or take it seriously. The iPhone also recommended different exercises I can do to help improve my steadiness.

Now I think that is valuable for a senior like me. I also think these new systems in cars are just as valuable. They pay for themselves the first time they help you avoid a collision.

Per the 1980 Citation. My Pontiac version, 1980 Pontiac Phoenix Sport never gave me problems. Also had the 1981 Buick version.

 

[Job]

You've referenced the human ability to see an image in multiple posts in this thread. Why?

There is no doubt that the best human driver is currently better than the best self-driving system in most circumstances - today. FSD doesn't need to learn, see, or understand an image as well as a human to eventually become safer and more capable. The people behind FSD will tweak the programming after each learning experience. Something went bad today? Ok, what do we need to tweak in the logic to keep that from happening next time? And so on. With each revision, there are less bad days.

People screw up a lot. Computers will do better.

Not just the ability to see an image, create the image. There is no other image than what the brain creates. A camera creates no image by itself. Think about it, what image is seen except by the brain? None. No videos, no pictures, nothing.
The argument a fighter pilot needs a computer is no different than a carpenter needs a hammer to pull out a nail. The tellers of such things forget the fighter pilot and carpenter are using tools. Of course humans who are built to walk on the ground can’t be a fighter jet themselves. The whole fighter is a tool. If we want to sit in the back seat and sleep while traveling in a car, the tool, take a taxi. Maybe when cars are controlled like people movers I referenced before, that’s ok, its on a track and still there somewhere are brains not computers watching over it.
What the original topic was is the Tesla braking for no reason. A human brain sees an essentially live image. It assesses everything that’s there. A camera does no thinking, or seeing, it responds to sensor pixel data. It’s a dead inert thing. A large plastic bag might as well be a moose, possibly.
That said I like front emergency braking, although it can throw warnings it shouldn’t. Because maybe I am trying to figure out how to use the rest of the garbage in new cars I take my eyes off the road. That’s all want it for. Or if I fall asleep which has never happened.

 

[ZeeOSix]

Horses were the original self driving AI ...

Have you ever ridden a horse? Get a spooky one, and it can do some crazy unexpected things, regarless of how good the rider is or how trained the horse is ... like a Tesla slamming the brakes on for no apparent reason. Riding a crazy horse is like AI going bonkers, and can be pretty dangerous.

 

[ZeeOSix]

Human has 16 years to develop the motor skill since birth before he or she can take a driving test. Self driving is "getting there", and usually good enough to be as good as a 16 year old new driver.

The way humans drive today, it should be more like 25 years old.

 

[PandaBear]

Have you ever ridden a horse? Get a spooky one, and it can do some crazy unexpected things, regarless of how good the rider is or how trained the horse is ... like a Tesla slamming the brakes on for no apparent reason. Riding a crazy horse is like AI going bonkers, and can be pretty dangerous.

Well, there's nothing that cannot be spooky occasionally, human, AI, horses, donkeys, orangutans.

 

[ZeeOSix]

Well, there's nothing that cannot be spooky occasionally, human, AI, horses, donkeys, orangutans.

I wouldn't want to be driving a vehicle with spooky AI or AP. Just like a spooky horse, you'd never know what it's capable of or when it might get "spooked".

 

[PandaBear]

I wouldn't want to be driving a vehicle with spooky AI or AP. Just like a spooky horse, you'd never know what it's capable of or when it might get "spooked".

Nor a spooked human driver?

 

[ZeeOSix]

Nor a spooked human driver?

I won't ride with spooked/bad drivers. I'm probably more prepared and have more foresight to make a defensive move from them than any current AI vehicle. I've ridden motorcycles for over 40 years on the streets (you can not day-dream, unless you want to be dead), and IMO that really hones the skills to be able to read the boneheads on the road and be prepared for a defensive/evasive move ... something no AI machine has the ability to do. All they can do is react to a situation ... they can not read and have foresight of the situations.

 

[OVERKILL]

I won't ride with spooked/bad drivers. I'm probably more prepared and have more foresight to make a defensive move from them than any current AI vehicle. I've ridden motorcycles for over 40 years on the streets (you can not day-dream, unless you want to be dead), and IMO that really hones the skills to be able to read the boneheads on the road and be prepared for a defensive/evasive move ... something no AI machine has the ability to do. All they can do is react to a situation ... they can not read and have foresight of the situations.

No, and they are likely not going to choose the rhubarb in a situation where you could end up dead (pile-up) vs just going off the road. There are so many situations that require driver intuition to know what the right course of action is, even if it would be the "wrong" course of action for AI.